How to Automate Webflow CMS Backups With Make.com in 2026

Why I Started Backing Up Every Webflow CMS Collection With Make.com After One Bad Tuesday

Last September a client's marketing intern accidentally bulk deleted 142 blog posts from their Webflow CMS. Webflow does not currently roll back CMS deletions. We rebuilt those 142 posts from the Google cache and screenshot archives. It took six days. I billed for none of it, because I had never put a backup system in place. That Tuesday I built one, and I have run it on every client site since.

Make.com sits in the middle of that backup system because it speaks Webflow's API natively and runs on a schedule I can trust. Make's May 2026 platform report listed Webflow as its third most connected app, behind only Google Sheets and Notion. I run weekly snapshots of every CMS collection across all 11 active client sites for a total cost of 23 dollars per month.

In this tutorial I will walk through the exact Make.com scenario I use, how the Webflow API rate limits affect the design, and how I store and restore the data when the worst case hits. This is the post I wish I had read in August 2025.

What Does a Webflow CMS Backup Actually Need to Capture in 2026?

A complete Webflow CMS backup captures every item in every collection, every field on every item, every published and draft state, every reference and multi reference, and every locale variant. Anything less leaves you with a partial restore that takes longer than rebuilding from scratch. The Webflow Data API exposes all of this, but you have to ask for it the right way.

Webflow's June 2026 API documentation lists 18 endpoints for CMS reads, and the ones that matter for a backup are list_collection_items, get_collection_details, and the locale variant endpoint. Together they let you reconstruct a collection from JSON in about 40 lines of Make.com modules. The cost is a few thousand API calls per run.

The trap I have seen Webflow practices fall into is backing up only the published items. If your client edits a draft and the live item never updated, your backup is wrong. Capture both states or the restore will fail silently when it matters.

How Do You Connect Make.com to Webflow Without Hitting Rate Limits?

Use a Webflow Site Token with read scope, set Make.com's concurrency to one, and add a five second delay between collection reads. The Webflow Data API allows 60 requests per minute per token as of the May 2026 limits update. A delay of five seconds keeps you well under the cap even when a collection has hundreds of items.

The token itself lives in your Webflow Workspace settings under Apps and Integrations. Generate a Site Token rather than a Workspace Token so the access stays scoped to a single site. Make.com stores the token as a connection that you can reuse across scenarios without re entering it. Rotate the token every 90 days as a habit.

For the actual scenario design, Make.com's iterator module is the piece that lets you walk through each collection one by one. Pair it with the array aggregator to assemble the full export before writing to storage. This pattern is what kept my early backup scenarios from blowing past the rate limit when I added new client sites.

How Do You Build the Make.com Scenario Step by Step?

The scenario starts with a scheduled trigger, fans out to a list of collections, walks through items in each, and writes the result to Google Drive. The full module count is seven, which keeps each run under Make.com's free tier limit if you only run weekly. I run mine every Saturday at 2 AM IST so it finishes before any client opens the Editor.

Module one is a schedule trigger. Module two calls the Webflow get_collection_list endpoint for the site. Module three iterates the collection IDs. Module four calls list_collection_items with limit 100 for each ID. Module five paginates through the offsets until the response is empty. Module six aggregates the items per collection. Module seven writes a single JSON file per collection to Google Drive.

The JSON file naming I use is site_slug_collection_slug_YYYYMMDD.json. That pattern lets me search the archive in one query and restore a single collection if I need to. I keep 26 weeks of snapshots, which is enough for any client situation I have seen.

How Do You Store the Backups So They Are Actually Restorable?

Store them as raw JSON in cold storage with a clear folder hierarchy. I use Google Drive folders organized by client and date, with a master index sheet that lists every snapshot and its row count. Raw JSON is the format the Webflow API will accept on the way back in, so the restore path is straightforward.

Forrester's April 2026 SaaS resilience study found that 68 percent of CMS rebuilds fail their first restore attempt because the backup format does not match the API write schema. Raw JSON exports straight from the read endpoint avoid this trap. The data shape on the way out matches the data shape on the way in.

I also keep an offsite copy in a Cloudflare R2 bucket because I do not want my entire backup history living on a single Google account. Cloudflare R2 charges no egress fees, which means restoring 26 weeks of data costs me nothing in transfer. R2 storage is 1.50 dollars per terabyte per month as of June 2026.

How Do You Restore a Webflow CMS Collection From a Make.com Backup?

Build a second Make.com scenario that reads the JSON file, walks through each item, and calls the Webflow create_collection_items endpoint in batches of three. The restore is slower than the backup because Webflow's write rate limit is tighter. Plan for about 30 minutes per 1,000 items on a normal token.

The detail that catches teams out is reference fields. If your Posts collection points at a Categories collection, restore the Categories first and capture the new item IDs. Then map the old reference IDs in the Posts backup to the new IDs before the restore call. Make.com's data store module is the easiest place to hold that mapping.

If you want the broader context on automation workflows that pair with backup discipline, my piece on Webflow Logic alternatives after the 2025 discontinuation walks through the Make.com versus Zapier versus n8n tradeoff for these kinds of scenarios.

How Often Should You Run a Webflow CMS Backup in 2026?

Weekly for most client sites, daily for any site with a content team larger than three people, and on demand before any major bulk edit. The frequency rule is that a backup should never be older than the time you would spend rebuilding by hand. For my Bengaluru fintech client, that interval is 24 hours.

Mailmodo's May 2026 small team automation survey found that 71 percent of small business sites running on Webflow had no scheduled backup of any kind. The same survey found that the 29 percent who did saved an average of 14 hours per incident when something went wrong. The math is one sided.

For the larger pattern of which workflows to automate first in a Webflow practice, my walkthrough on building Webflow form automation with Logic and Make.com covers the sibling scenario you will probably want to run alongside the backup.

What Are the Common Failure Modes I Have Hit Running This For Two Years?

The three failures I see most are token expiry, rate limit overrun on big collections, and silent reference field corruption on restore. Token expiry is the easiest to fix: set a calendar reminder every 80 days. Rate limit overrun shows up when you add a new client without adjusting the global concurrency setting. Reference corruption shows up when you restore one collection without remapping IDs.

The fourth failure, which I hit once in November 2025, is a Make.com scenario silently turning off after a billing event. Make.com pauses scenarios on payment failure without warning, and the backup just stops running. I now monitor scenario run health with a Google Sheet that flags any week missing a row.

None of these failures are dramatic. They are the kind of slow degradation that shows up the week your client needs the backup most. Build the monitoring first, then trust the automation.

How to Ship Your First Webflow CMS Backup Scenario This Week

Generate a Webflow Site Token, sign up for Make.com's free tier, and build the seven module scenario described above. Run it once manually, verify the JSON output looks right in Google Drive, and then schedule it for weekly. Total setup time on your first site is about 90 minutes. Each additional site clones in 15.

If you want to compare this against a simpler backup option before you commit, the Webflow Editor still allows manual collection exports as CSV. CSV is fine for emergencies but loses field type information, so a Make.com restore from CSV needs more cleanup. I recommend JSON for any client site you intend to support long term.

If you want help wiring this up on your own Webflow stack, I am happy to walk through it. Let's chat.